RFR (S): 8188919: Generalize GC thread suspend/resume at safepoints

[Erik Österlund]

Hi Per,

Thanks for the review.

/Erik

On 2017-10-17 09:11, Per Liden wrote:
> Looks good!
>
> /Per
>
> On 2017-10-13 13:58, Erik Österlund wrote:
>> Hi Roman,
>>
>> I waited for your changes to be upstreamed and rebased on top of that
>> as you suggested.
>>
>> New webrev:
>> http://cr.openjdk.java.net/~eosterlund/8188919/webrev.01/
>>
>> Thanks,
>> /Erik
>>
>> On mån, 2017-10-09 at 17:46 +0200, Roman Kennke wrote:
>>> Am 09.10.2017 um 17:35 schrieb Erik Österlund:
>>>>
>>>> Hi,
>>>>
>>>> When we synchronize/unsynchronize safepoints, we make sure to
>>>> perform
>>>> corresponding suspend/resume calls to GC-specific concurrent
>>>> workers
>>>> that may interfere with the safepoint. This is currently done by
>>>> dumping GC-specific callbacks right in
>>>> SafepointSynchronize::begin()
>>>> and SafepointSynchronize::end(). The purpose with this cleanup is
>>>> to
>>>> hide this behind a general interface on CollectedHeap instead.
>>>>
>>>> Webrev:
>>>> http://cr.openjdk.java.net/~eosterlund/8188919/webrev.00/
>>>>
>>>> Bug:
>>>> https://bugs.openjdk.java.net/browse/JDK-8188919
>>>>
>>>> Thanks,
>>>> /Erik
>>> I like it.
>>>
>>> It's going to conflict with this almost-pushed changeset though:
>>>
>>> https://bugs.openjdk.java.net/browse/JDK-8179387
>>>
>>> In fact, I think it's be cleaner to base your patch on this one.
>>>
>>> Roman
>>>
>>>